From 22533dc739a4471c8dcfd4a1daf1135d042e5c8d Mon Sep 17 00:00:00 2001
From: Kevin Pham <kevin@deoxy.net>
Date: Sat, 9 Dec 2023 08:37:23 -0600
Subject: [PATCH] improve ruleset unmarshalling behavior if there isnt a top
 level rule key

---
 handlers/api_modifiers_structdef.gen.go        |  9 +++++++++
 handlers/proxy.go                              |  4 ++--
 .../patch_dynamic_resource_urls.go             | 13 +++++++++++++
 proxychain/ruleset/rule_resmod_types.gen.go    |  4 ++++
 proxychain/ruleset/ruleset.go                  | 18 ++++++++++++++----
 ruleset_v2.yaml                                | 16 ++++++++++++++++
 6 files changed, 58 insertions(+), 6 deletions(-)

diff --git a/handlers/api_modifiers_structdef.gen.go b/handlers/api_modifiers_structdef.gen.go
index 15fff3e..bc4caff 100644
--- a/handlers/api_modifiers_structdef.gen.go
+++ b/handlers/api_modifiers_structdef.gen.go
@@ -513,6 +513,15 @@ var AllMods Modifiers = Modifiers{
     {Name: "val", Type: "string"},
   },
 },
+{
+  Name:        "ModifyIncomingScriptsWithRegex",
+  Description: "ModifyIncomingScriptsWithRegex modifies all incoming javascript (application/javascript and inline <script> in text/html) using a regex match and replacement.",
+  CodeEditLink: "https://github.com/everywall/ladder/edit/origin/proxy_v2/proxychain/responsemodifiers/modify_incoming_scripts_with_regex.go",
+  Params: []Param{
+    {Name: "matchRegex", Type: "string"},
+    {Name: "replacement", Type: "string"},
+  },
+},
 {
   Name:        "SetResponseHeader",
   Description: "SetResponseHeader modifies response headers from the upstream server",
diff --git a/handlers/proxy.go b/handlers/proxy.go
index d08fe30..2db8827 100644
--- a/handlers/proxy.go
+++ b/handlers/proxy.go
@@ -45,7 +45,7 @@ func NewProxySiteHandler(opts *ProxyOptions) fiber.Handler {
 			).
 			AddResponseModifications(
 				tx.ForwardResponseHeaders(),
-				//tx.BlockThirdPartyScripts(),
+				tx.BlockThirdPartyScripts(),
 				tx.DeleteIncomingCookies(),
 				tx.DeleteLocalStorageData(),
 				tx.DeleteSessionStorageData(),
@@ -54,7 +54,7 @@ func NewProxySiteHandler(opts *ProxyOptions) fiber.Handler {
 				tx.RewriteHTMLResourceURLs(),
 				tx.PatchDynamicResourceURLs(),
 				tx.PatchTrackerScripts(),
-				tx.BlockElementRemoval(".article-content"), // techcrunch
+				//tx.BlockElementRemoval(".article-content"), // techcrunch
 				//tx.BlockElementRemoval(".available-content"), // substack
 			// tx.SetContentSecurityPolicy("default-src * 'unsafe-inline' 'unsafe-eval' data: blob:;"),
 			)
diff --git a/proxychain/responsemodifiers/patch_dynamic_resource_urls.go b/proxychain/responsemodifiers/patch_dynamic_resource_urls.go
index b418b54..cb8d354 100644
--- a/proxychain/responsemodifiers/patch_dynamic_resource_urls.go
+++ b/proxychain/responsemodifiers/patch_dynamic_resource_urls.go
@@ -52,6 +52,19 @@ func PatchDynamicResourceURLs() proxychain.ResponseModification {
 		htmlRewriter := rewriters.NewHTMLRewriter(chain.Response.Body, rr)
 		chain.Response.Body = htmlRewriter
 
+		// window.location
+		/*
+				spoofedLocationAPI := fmt.Sprintf(`{href:"%s", origin:"%s", pathname:"%s", protocol:"%s:", port:"%s"}`,
+					reqURL.String(), reqURL.Host,
+					reqURL.Path, reqURL.Scheme, reqURL.Port())
+			spoofedLocationAPI := fmt.Sprintf(`{origin: "%s"}`, reqURL.Host)
+			fmt.Println(spoofedLocationAPI)
+
+			chain.AddOnceResponseModifications(
+				ModifyIncomingScriptsWithRegex(`window\.location`, spoofedLocationAPI),
+			)
+		*/
+
 		return nil
 	}
 }
diff --git a/proxychain/ruleset/rule_resmod_types.gen.go b/proxychain/ruleset/rule_resmod_types.gen.go
index bce41d3..59bfe84 100644
--- a/proxychain/ruleset/rule_resmod_types.gen.go
+++ b/proxychain/ruleset/rule_resmod_types.gen.go
@@ -84,6 +84,10 @@ func init() {
     return tx.SetIncomingCookie(params[0], params[1])
   }
 
+  rsmModMap["ModifyIncomingScriptsWithRegex"] = func(params ...string) proxychain.ResponseModification {
+    return tx.ModifyIncomingScriptsWithRegex(params[0], params[1])
+  }
+
   rsmModMap["SetResponseHeader"] = func(params ...string) proxychain.ResponseModification {
     return tx.SetResponseHeader(params[0], params[1])
   }
diff --git a/proxychain/ruleset/ruleset.go b/proxychain/ruleset/ruleset.go
index 7f28d2b..c3cd098 100644
--- a/proxychain/ruleset/ruleset.go
+++ b/proxychain/ruleset/ruleset.go
@@ -33,14 +33,24 @@ func (rs *Ruleset) UnmarshalYAML(unmarshal func(interface{}) error) error {
 	type AuxRuleset struct {
 		Rules []Rule `yaml:"rules"`
 	}
-	yr := &AuxRuleset{}
+	yamlRuleset := &AuxRuleset{}
 
-	if err := unmarshal(&yr); err != nil {
-		return err
+	if err := unmarshal(&yamlRuleset); err != nil {
+		// if there is no top-level rule key, we'll try to unmarshal as if it is just a bare rule
+		recovered := false
+		yamlRule := &Rule{}
+		if err := unmarshal(&yamlRule); err != nil {
+			yamlRuleset.Rules = append(yamlRuleset.Rules, *yamlRule)
+			recovered = true
+		}
+
+		if !recovered {
+			return err
+		}
 	}
 
 	rs._rulemap = make(map[string]*Rule)
-	rs.Rules = yr.Rules
+	rs.Rules = yamlRuleset.Rules
 
 	// create a map of pointers to rules loaded above based on domain string keys
 	// this way we don't have two copies of the rule in ruleset
diff --git a/ruleset_v2.yaml b/ruleset_v2.yaml
index 0eea0ec..6875d64 100644
--- a/ruleset_v2.yaml
+++ b/ruleset_v2.yaml
@@ -15,3 +15,19 @@ rules:
     requestmodifications:
     - name: ForwardRequestHeaders
       params: []
+
+  - domains:
+    - quantamagzine.org
+    responsemodifications:
+    - name: BlockElementRemoval
+      params:
+      - "#postContent"
+
+  - domains:
+    - techcrunch.com
+    responsemodifications:
+    - name: ModifyIncomingScriptsWithRegex
+      params:
+      - "window\\.location"
+      - |
+        {origin: "techcrunch.com"}